Moore’s Law: 50 Years and Beyond

September 08, 2015 by Mark Gerasimas

Examining Moore's Law in its 50th Year and how the initial premise may be approaching exhaustion, but the spirit of this observation continues.

In the April, 1965 edition of Electronics magazine, Gordon Moore made an ambitious but not altogether extraordinary observation about the trend of component size on semiconductor wafers.  In the article titled “Cramming More Components Onto Integrated Circuits," the co-founder of Fairchild Semiconductor stated that the number of components "crammed" onto an integrated circuit could easily double every year for the next ten years.  His analysis held true until 1975 when he refined his observation to a doubling of components per circuit (or conversely, halving of component size) to once every two years instead of one.  This was due to Moore's observation that an initially increased rate was aided by the optimization of components required to store a single bit of data, thereby reducing the number of components required to accomplish the same task.  This simple observation set the tone of electronic innovation for the next 50 years. Thus, Moore's Law was born!

Oughta Be a Law

Ok, Moore's Law isn't exactly a law; it's more of an observation that has held true for a period of approximately 60 years and serves as a production model by which semiconductor producers base their production timelines.  The goal for these developers has consistently been to cut component size (namely transistors) in half every two years while keeping integrated circuit cost at a minimum.  The end result can be manipulated to meet the consumer need.  The three typical theoretical cases are below.

- More components on the same size integrated circuit gives you a more powerful device in the same size package for the same price.

- The same number of components on a smaller integrated circuit gives you the same processing power on a smaller device for the same price.

- The same sized device with the same processing power allows for a cheaper price.

The Benchmark

Moore's Law established the standard by which the semiconductor industry measures three metrics: circuit cost, circuit speed, transistor size.  When these three characteristics are optimized, we as consumers see the result as smaller components with improved processing power at a cheaper price.  The effort to optimize circuit components has been led by Intel Corporation (also founded by Gordon Moore) since 1968.  Intel's research and development teams have cut transistor size to keep pace with Moore's Law.  Advancements like these allowed the electronics industry to drive performance into ever more convenient packages.  

For example, the first general purpose electronic computer in the world (Electronic Numerical Integrator And Computer, or ENIAC for short), sponsored by the US Army and designed at the University of Pennsylvania in 1943, was able to execute simple mathematical programs (similar to Texas Instruments' Ti-83 calculator) coded into it through a series of switches and vacuum tubes.  However, the ENIAC weighed 27 tons and occupied 1800ft2.  The Ti-83 was released in 1996, weighs only 0.5 pounds, and occupies just under 25in3.  Of note, the Army sponsored the development of ENIAC to compute artillery firing solution tables for combat, at the low price of $6,000,000 in today's dollars.  The Texas Instruments developed the Ti-83 to torture high school and college calculus students during class and feed their addiction to games such as Block Dude and Drug Wars, all for the low price of $88 with free shipping.  Today's smart phones have significantly more capability in a smaller sized package.  Clearly, the industry has kept pace.

The End of the Road?

With the 50th Anniversary of Moore's Law, technical critics have stated that this critical observation is outdated and that the law is dead.  Intel has admitted that the pace to halve the size of its 14nm transistor width technology is likely to take more like two and a half years, and skeptics are doubtful of how much smaller they can make their tri-gate transistor.  Gordon Moore himself has even stated that he saw his law "dying here in the next decade or so."  This is due to the obstacles presented by working with silicon on the order of 5nm or less.  At a minimum, it is safe to say that the end of progress in silicon based semiconductor transistor size is at hand.  

However, while the letter of Moore's Law is coming to an end, the drive and spirit of innovation it instilled in the industry will continue its trend.  The drive of researchers to develop more powerful devices in smaller packages for a lower cost will continue through research into other techniques, materials, and configurations.  The massive migration to cloud computing has already shown promise through tremendous computing power accessed through ultra-fast connectivity on your personal device.  Perhaps Mr. Moore will have a further prediction for us in the next ten years.  As it's always been with Moore's Law, only time will tell.